In addition to payload size, you should also measure payload complexity. It's a lot easier and faster to transform a single 10MB string than it is to transform a 10K array containing 100 complex structures.

But I have to ask why you're planning to compare the two. The performance difference should be relatively minor. I expect that RPC/encoded will be somewhat faster if you're trying to pass complex object graphs, but doing so will give you enormous interoperability headaches. Your interface should pass flat document structures (simple types, structures of simple types, and arrays). Period. Besides -- to be WS-I conformant you must not use SOAP encoding.

You should see no difference in performance at all between RPC/Literal and Document/Literal -- assuming that you're asking the SOAP engine to perform the serialization. (Note that Axis and .NET don't support RPC/Literal.)

Perhaps you're looking to compare Axis serialization versus program-based serialization or versus a third party binding framework (JAXB, JiBX, Castor, XMLBeans, etc)? That might be interesting.

Keep in mind, though, that your results apply only to the specific products you're testing. You are likely to get very different results if you use Axis versus Sun JAX-RPC versus webMethods Glue versus Systinet WASP versus BEA WebLogic versus IBM WebSphere versus Novell exteNd versus IONA Artix versus Microsoft .NET, etc.

Anne


At 02:40 PM 3/30/2004, you wrote:
Hello All,

I am doing a performance evaluation between Document and RPC style web services. Below are two of the criteria I would be evaluating.....

i)  Requests/Sec served (Vs) Payload size
ii) Time taken to process a single request (Vs) Variable Payload Size

Can anyone think of any better criteria to evaluate?

Thanks a lot,
Vivek

_________________________________________________________________
Get reliable access on MSN 9 Dial-up. 3 months for the price of 1! (Limited-time offer) http://join.msn.com/?page=dept/dialup&pgmarket=en-us&ST=1/go/onm00200361ave/direct/01/

~~~~~~~~~~~~~~~~~~
Anne Thomas Manes
VP & Research Director
Burton Group




Reply via email to